Skip to content
体验新版
项目
组织
正在加载...
登录
切换导航
打开侧边栏
Greenplum
Opencv
提交
1c9f956b
O
Opencv
项目概览
Greenplum
/
Opencv
大约 1 年 前同步成功
通知
7
Star
0
Fork
0
代码
文件
提交
分支
Tags
贡献者
分支图
Diff
Issue
0
列表
看板
标记
里程碑
合并请求
0
DevOps
流水线
流水线任务
计划
Wiki
0
Wiki
分析
仓库
DevOps
项目成员
Pages
O
Opencv
项目概览
项目概览
详情
发布
仓库
仓库
文件
提交
分支
标签
贡献者
分支图
比较
Issue
0
Issue
0
列表
看板
标记
里程碑
合并请求
0
合并请求
0
Pages
DevOps
DevOps
流水线
流水线任务
计划
分析
分析
仓库分析
DevOps
Wiki
0
Wiki
成员
成员
收起侧边栏
关闭侧边栏
动态
分支图
创建新Issue
流水线任务
提交
Issue看板
体验新版 GitCode,发现更多精彩内容 >>
提交
1c9f956b
编写于
12月 04, 2014
作者:
Y
Yan Wang
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
Remove unnecesary operations and calculations in loop body.
Signed-off-by:
N
Yan Wang
<
yan.wang@linux.intel.com
>
上级
be5c9103
变更
1
隐藏空白更改
内联
并排
Showing
1 changed file
with
27 addition
and
15 deletion
+27
-15
modules/features2d/src/opencl/brute_force_match.cl
modules/features2d/src/opencl/brute_force_match.cl
+27
-15
未找到文件。
modules/features2d/src/opencl/brute_force_match.cl
浏览文件 @
1c9f956b
...
...
@@ -210,6 +210,8 @@ __kernel void BruteForceMatch_Match(
}
#
else
__local
value_type
*s_train
=
(
__local
value_type
*
)
sharebuffer
+
BLOCK_SIZE_ODD
*
BLOCK_SIZE
;
const
int
s_query_i
=
mad24
(
BLOCK_SIZE_ODD,
lidy,
lidx
)
;
const
int
s_train_i
=
mad24
(
BLOCK_SIZE_ODD,
lidx,
lidy
)
;
#
endif
float
myBestDistance
=
MAX_FLOAT
;
...
...
@@ -242,13 +244,15 @@ __kernel void BruteForceMatch_Match(
{
const
int
loadx
=
mad24
(
i,
BLOCK_SIZE,
lidx
)
;
//load
query
and
train
into
local
memory
s_query[mad24
(
BLOCK_SIZE_ODD,
lidy,
lidx
)
]
=
0
;
s_train[mad24
(
BLOCK_SIZE_ODD,
lidx,
lidy
)
]
=
0
;
if
(
loadx
<
query_cols
)
{
s_query[mad24
(
BLOCK_SIZE_ODD,
lidy,
lidx
)
]
=
query_vec[loadx]
;
s_train[mad24
(
BLOCK_SIZE_ODD,
lidx,
lidy
)
]
=
train_vec[loadx]
;
s_query[s_query_i]
=
query_vec[loadx]
;
s_train[s_train_i]
=
train_vec[loadx]
;
}
else
{
s_query[s_query_i]
=
0
;
s_train[s_train_i]
=
0
;
}
barrier
(
CLK_LOCAL_MEM_FENCE
)
;
...
...
@@ -337,18 +341,22 @@ __kernel void BruteForceMatch_RadiusMatch(
__local
value_type
*s_train
=
(
__local
value_type
*
)
sharebuffer
+
BLOCK_SIZE_ODD
*
BLOCK_SIZE
;
result_type
result
=
0
;
const
int
s_query_i
=
mad24
(
BLOCK_SIZE_ODD,
lidy,
lidx
)
;
const
int
s_train_i
=
mad24
(
BLOCK_SIZE_ODD,
lidx,
lidy
)
;
for
(
int
i
=
0
; i < (query_cols + BLOCK_SIZE - 1) / BLOCK_SIZE ; ++i)
{
//load
a
BLOCK_SIZE
*
BLOCK_SIZE
block
into
local
train.
const
int
loadx
=
mad24
(
BLOCK_SIZE,
i,
lidx
)
;
s_query[mad24
(
BLOCK_SIZE_ODD,
lidy,
lidx
)
]
=
0
;
s_train[mad24
(
BLOCK_SIZE_ODD,
lidx,
lidy
)
]
=
0
;
if
(
loadx
<
query_cols
)
{
s_query[mad24
(
BLOCK_SIZE_ODD,
lidy,
lidx
)
]
=
query_vec[loadx]
;
s_train[mad24
(
BLOCK_SIZE_ODD,
lidx,
lidy
)
]
=
train_vec[loadx]
;
s_query[s_query_i]
=
query_vec[loadx]
;
s_train[s_train_i]
=
train_vec[loadx]
;
}
else
{
s_query[s_query_i]
=
0
;
s_train[s_train_i]
=
0
;
}
//synchronize
to
make
sure
each
elem
for
reduceIteration
in
share
memory
is
written
already.
...
...
@@ -405,6 +413,8 @@ __kernel void BruteForceMatch_knnMatch(
}
#
else
__local
value_type
*s_train
=
(
__local
value_type
*
)
sharebuffer
+
BLOCK_SIZE_ODD
*
BLOCK_SIZE
;
const
int
s_query_i
=
mad24
(
BLOCK_SIZE_ODD,
lidy,
lidx
)
;
const
int
s_train_i
=
mad24
(
BLOCK_SIZE_ODD,
lidx,
lidy
)
;
#
endif
float
myBestDistance1
=
MAX_FLOAT
;
...
...
@@ -438,13 +448,15 @@ __kernel void BruteForceMatch_knnMatch(
{
const
int
loadx
=
mad24
(
BLOCK_SIZE,
i,
lidx
)
;
//load
query
and
train
into
local
memory
s_query[mad24
(
BLOCK_SIZE_ODD,
lidy,
lidx
)
]
=
0
;
s_train[mad24
(
BLOCK_SIZE_ODD,
lidx,
lidy
)
]
=
0
;
if
(
loadx
<
query_cols
)
{
s_query[mad24
(
BLOCK_SIZE_ODD,
lidy,
lidx
)
]
=
query_vec[loadx]
;
s_train[mad24
(
BLOCK_SIZE_ODD,
lidx,
lidy
)
]
=
train_vec[loadx]
;
s_query[s_query_i]
=
query_vec[loadx]
;
s_train[s_train_i]
=
train_vec[loadx]
;
}
else
{
s_query[s_query_i]
=
0
;
s_train[s_train_i]
=
0
;
}
barrier
(
CLK_LOCAL_MEM_FENCE
)
;
...
...
编辑
预览
Markdown
is supported
0%
请重试
或
添加新附件
.
添加附件
取消
You are about to add
0
people
to the discussion. Proceed with caution.
先完成此消息的编辑!
取消
想要评论请
注册
或
登录